https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46447
Jonathan Wakely <redi at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Last reconfirmed| |2020-12-16
Status|UNCONFIRMED |ASSIGNED
Target Milestone|--- |11.0
Assignee|bkoz at gcc dot gnu.org |redi at gcc dot gnu.org
Ever confirmed|0 |1
--- Comment #4 from Jonathan Wakely <redi at gcc dot gnu.org> ---
with GCC 8 and 9 the results are very consistent on x86_64: atomic_flag and
atomic_uchar and atomic_int all have identical performance, but clearing is ~4
times slower than setting. The spin_mutex test is very slightly slower than the
sum of atomic_flag::test_and_set() and atomic_flag::clear() (which makes sense,
since that's what it does).
With GCC 10 and trunk, clear is now as fast as set.